On Thu, Feb 06, 2025 at 02:27:16PM +0100, Darrick J. Wong wrote:
> On Thu, Feb 06, 2025 at 07:00:55PM +0000, da.gomez@xxxxxxxxxx wrote:
> > From: Daniel Gomez <da.gomez@xxxxxxxxxxx>
> >
> > In patch [1] ("bdev: use bdev_io_min() for statx block size"), block
> > devices will now report their preferred minimum I/O size for optimal
> > performance in the stx_blksize field of the statx data structure. This
> > change updates the current default 4 KiB block size for all devices
> > reporting a minimum I/O larger than 4 KiB, opting instead to query for
> > its advertised minimum I/O value in the statx data struct.
> >
> > [1]:
> > https://lore.kernel.org/all/20250204231209.429356-9-mcgrof@xxxxxxxxxx/
>
> This isn't even upstream yet...
>
> > Signed-off-by: Daniel Gomez <da.gomez@xxxxxxxxxxx>
> > ---
> > Set MIN-IO from statx as the default filesystem fundamental block size.
> > This ensures that, for devices reporting values within the supported
> > XFS block size range, we do not incur in RMW. If the MIN-IO reported
> > value is lower than the current default of 4 KiB, then 4 KiB will be
> > used instead.
>
> I don't think this is a good idea -- assuming you mean the same MIN-IO
> as what lsblk puts out:
This is just about matching the values in code and documentation across all
layers (to guarantee writes do no incur in RMW when possible and supported by
the fs): minimum_io_size (block layer) -> stx_blksize (statx) -> lsblk MIN-IO
(minimum I/ O size) -> Filesystem fundamental block size (mkfs.xfs -b size).
* MIN-IO is the minimum I/O size in lsblk [1] which should be the queue-sysfs
minimum_io_size [2] [3] ("This is the smallest preferred IO size reported by the
device").
* From statx [4] manual (and kernel statx data struct description), stx_blksize is
'The "preferred" block size for efficient filesystem I/O (Writing to a file in
smaller chunks may cause an inefficient read-modify-rewrite.)'
[1] https://github.com/util-linux/util-linux/blob/master/misc-utils/lsblk.c#L199
[2] minimum_io_size: https://www.kernel.org/doc/Documentation/block/queue-sysfs.txt
[3] https://www.kernel.org/doc/Documentation/ABI/stable/sysfs-block
What: /sys/block/<disk>/queue/minimum_io_size
Date: April 2009
Contact: Martin K. Petersen <martin.petersen@xxxxxxxxxx>
Description:
[RO] Storage devices may report a granularity or preferred
minimum I/O size which is the smallest request the device can
perform without incurring a performance penalty. For disk
drives this is often the physical block size. For RAID arrays
it is often the stripe chunk size. A properly aligned multiple
of minimum_io_size is the preferred request size for workloads
where a high number of I/O operations is desired.
[4] https://git.kernel.org/pub/scm/docs/man-pages/man-pages.git/tree/man/man2/statx.2?id=master#n369
kernel: __u32 stx_blksize; /* Preferred general I/O size [uncond] */
> nvme1n1 512
> └─md0 524288
> └─node0.raid 524288
> └─node0_raid-storage 524288
Is the MIN-IO correctly reported in RAID arrays here? I guess it should match
the stripe chunk size as per description above?
